I have two studies (same patient etc ) that have two different StudyInstanceUID.
But all the SOPInstanceUID in the two studies are the same.
The images are besides this exactly the same.

Is this allowed?
If not, where is that rule specified.

Thanks and Kind regards

On Thursday, November 17, 2022 at 11:19:07 PM UTC+1, tomas moberg wrote:
> I have two studies (same patient etc ) that have two different StudyInstanceUID.
> But all the SOPInstanceUID in the two studies are the same.
> The images are besides this exactly the same.
>
> Is this allowed?

No.

> If not, where is that rule specified.

I would point to:

* https://dicom.nema.org/medical/dicom/current/output/chtml/part05/chapter_9.html

You should be able to C-GET using a SOPInstanceUID (Unique Key Attribute), so it must return a unique instance in this case.

Mathieu Malaterre schrieb am Freitag, 18. November 2022 um 09:49:48 UTC+1:
> On Thursday, November 17, 2022 at 11:19:07 PM UTC+1, tomas moberg wrote:
> > I have two studies (same patient etc ) that have two different StudyInstanceUID.
> > But all the SOPInstanceUID in the two studies are the same.
> > The images are besides this exactly the same.
> >
> > Is this allowed?
> No.
> > If not, where is that rule specified.
> I would point to:
>
> * https://dicom.nema.org/medical/dicom/current/output/chtml/part05/chapter_9.html
>
> You should be able to C-GET using a SOPInstanceUID (Unique Key Attribute), so it must return a unique instance in this case.

This happens if a PACS consolidates study information after acquisition. Most probably one StudyInstanceUID is modality generated, while the other is RIS generated. You can look up the OIDs at http://oid-info.com/ to prove that.

An institution has to make sure that the images they create are unique by defining a pipeline that all images go through after acquision and before releasing them into the wild.

fredag 18 november 2022 kl. 09:49:48 UTC+1 skrev Mathieu Malaterre:
> On Thursday, November 17, 2022 at 11:19:07 PM UTC+1, tomas moberg wrote:
> > I have two studies (same patient etc ) that have two different StudyInstanceUID.
> > But all the SOPInstanceUID in the two studies are the same.
> > The images are besides this exactly the same.
> >
> > Is this allowed?
> No.
> > If not, where is that rule specified.
> I would point to:
>
> * https://dicom.nema.org/medical/dicom/current/output/chtml/part05/chapter_9.html
>
> You should be able to C-GET using a SOPInstanceUID (Unique Key Attribute), so it must return a unique instance in this case.

Thanks.
But only if the Dicom Conformance Statement say you support c-get. This one is not supporting it.

This is a PACS where You can copy things in a unusual matter. Like adding the Topogram in a CT in to every series in that study.

The behaviour is crazy and I try to find a argument that I can make against it.
But I have not found it in the DICOM standard yet. I dont think it is in the chapter You linked to.

Kind regards

fredag 18 november 2022 kl. 18:28:34 UTC+1 skrev Sebastian Meyer:
> Mathieu Malaterre schrieb am Freitag, 18. November 2022 um 09:49:48 UTC+1:
> > On Thursday, November 17, 2022 at 11:19:07 PM UTC+1, tomas moberg wrote:
> > > I have two studies (same patient etc ) that have two different StudyInstanceUID.
> > > But all the SOPInstanceUID in the two studies are the same.
> > > The images are besides this exactly the same.
> > >
> > > Is this allowed?
> > No.
> > > If not, where is that rule specified.
> > I would point to:
> >
> > * https://dicom.nema.org/medical/dicom/current/output/chtml/part05/chapter_9.html
> >
> > You should be able to C-GET using a SOPInstanceUID (Unique Key Attribute), so it must return a unique instance in this case.
> This happens if a PACS consolidates study information after acquisition. Most probably one StudyInstanceUID is modality generated, while the other is RIS generated. You can look up the OIDs at http://oid-info.com/ to prove that.
>
> An institution has to make sure that the images they create are unique by defining a pipeline that all images go through after acquision and before releasing them into the wild.

Well, the image pixel data is unique with correspoding SOPInstanceUID. Its just the SeriesInstanceUID and the StudyInstanceUID that is cahnged.
I have seen RIS/Modlaity mixup with StudyInstanceUID, but this is not it. It is a PACS that allow you to copy objects within the same study and make copies with different StudyInstanceUID.
I agree with Your conclusion. But i can not find any sopport in the DICOM standard.
Kind regards

UIDs uniquely identify entities such as Studies, Series, SOP Instances, etc.

The relationship between such entities is explicitly described in the DICOM Information Model in general and specific to Composite IODs:

https://dicom.nema.org/medical/dicom/current/output/chtml/part03/chapter_7.html
https://dicom.nema.org/medical/dicom/current/output/chtml/part03/chapter_A.html#sect_A.1.2

Instances are contained within a single Series, Series are contained within a single Study, etc.

If that is not enough, it is even stated explicitly: "An image is related to a single Series within a single Study":

https://dicom.nema.org/medical/dicom/current/output/chtml/part03/chapter_A.html#para_934cbbbe-a76e-447d-9bac-f936b7ec791f